Wave Disturbances over the China Continent and the Eastern China Sea in February 1968 *

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 11 Wave Disturbances over the China Continent and the Eastern China Sea in February 1968 * By Tsuyoshi Nitta Department of Meteorology, University of California, Los Angeles Masatoshi Nanbu Department of Astronomy and Earth Sciences, Tokyo Gakugei University, Tokyo and Masanori Yoshizaki Geophysical Institute, Tokyo University, Tokyo (Manuscript received 18 July 1972, in revised form 12 January 1973) Abstract Wave disturbances appeared in the China continent and the Eastern China Sea in February 1968 are studied by the spectral method and the dynamical computations of the vertical velocity and the diabatic heating. Two types of disturbances are obtained. The one is the wave with 4- to 5-day period and the other is the wave with 1.5-to 2-day period. The disturbances with 4- to 5-day period exist in the regions to the south of 32*N where the southern jet currents flow. These disturbances propagate from west to east with the wavelength about 4,000 *5,000 km and their amplitude becomes large in the Eastern China Sea. The vartical structure of the disturbance is similar to that of a baroclinic unstable wave. Some disturbances of these waves develop suddenly as going over the Eastern China Sea. The diaba tic heating and the coupling between the lower tropospheric disturbances and the troughs in the upper troposphere may play an important role in the extreme development of the disturbances. Short-period disturbances with 1.5- to 2-day period exist in the lower troposphere below 600mb in the southern part of the Eastern China Sea especially during the period from 1 to 10 February. These disturbances have the wavelength about 2,000 km and their vertical structure is also similar to that of the baroclinic wave, i.e., the trough axes tilts slightly westward with height, warm air covers the region on and the east of the trough and upward motion occurs to the east of the trough. These medium-scale disturbances can be seen in the synoptic patterns. The Richardson number of the mean field is order of unity and these medium-scale disturbances are the baroclinic waves developing in the atmosphere of low Richardson number as suggested theoretically by Gambo (1970, a,b) and Tokioka (1970). 1. Introduction It has been well known that many disturbances appear in the Eastern China Sea especially during the season from February to April and sometimes develop into strong cyclones. The marked development of the disturbances in this area is considered to be closely related to the energy supply from the sea surface. Recently Ninomiya (1972) examined the heat energy budget in this area in February 1968. The amount of * This study is made as a part of the Air-Mass Transformation Experiment (AMTEX) of Japan. Division of Meteorology, Contribution No. ** On leave from Geophysical Institute, Tokyo University, Tokyo. the sensible heat supply and the evaporation were estimated to be 3001y/day and 10 mm/day respectively. He also found the disturbances with a period about 4 days in this area. Synoptic analyses about these disturbances have been reported by many authors but the dynamical structure of the disturbances and the mechanism of development have not been studied fully. In recent years medium-scale disturbances have collected attention of many meteorologists. The stability of the classical Norwegian polar front model has been investigated by Eliasen (1960) and Orlanski (1968). Orlanski (1968) found an unstable wave corresponding to a medium-scale disturbance which develops due to both effects of baroclinic instability and Helmnoltz instability

12 Journal of the Meteorological Society of Japan Vol. 51, No. 1 when the Richardson number (Ri) is of the order of unity and the Rossby number (Ro) is smaller than unity. Gambo (1970, a, b) examined the stability of medium-scale disturbances without considering a frontal discontinuity by using a model with uniform vertical shear. He showed in his numerical experiment that the characteristic features of disturbances are quite different according to the value of the Richardson number, i.e., whether Ri is larger than one or not. Later, Tokioka (1970, 1971) studied this problem more exactly on the range of small Richardson number. In comparison with the theoretical studies, there have been few data analyses about the mediumscale disturbances in the real atmosphere except the studies of Matsumoto, Yoshizumi and Takeuchi (1970). They obtained the small scale cyclone with a wavelength about 1,000 km along the "Baiu front". In the present paper we shall investigate characteristic features and structures of wave disturbances existing over the China continent and the Eastern China Sea in February 1968. 2. Data and method of analysis The data used in this study are upper air observations over the China continent and the Eastern China Sea in February 1968. The data at almost all stations are the twice daily observations and the wind data at Ishigakijima, Naze, Minamidaitojima and Kagoshima are available at four times in a day. The locations of the upperair stations are shown in Fig. 1. Wind, temperature, dew-point temperature and geopotential height data are available at stations in the China continent at the surface and constant pressure levels of 1,000, 850, 700, 500, 400, 300, 200 and 100mb. There are many missing data at 1,000mb and upper levels above 500mb. At stations in the Eastern China Sea and in the Japan Islands the data are available at the surface and the constant pressure levels of 1,000, 900, 850, 700, 600, 500, 400, 350, 300, 250, and 200mb. The enclosed areas I. II, III and IV in Fig. 1 are used in computation of vertical velocity and heat source. In order to examine period, wavelength and vertical structure of wave disturbances, we apply power spectral analysis to the time series data. Although the total period of the data are 29 days, we can detect characteristic feature of wave disturbances with the period shorter than 10 days. When we obtain the phase difference between two parameters from the cross spectrum, we adopt the results whose coherence is larger than a critical coherence corresponding to 95 probability level. The details of the spectral analysis were given in Maruyama (1968). We used a simple band-pass filter to pick up wave disturbances under consideration. For the wave disturbances with period about 4 days, we subtracted 4.5-day running mean from 1.5-day running mean of the original time series data, i.e., where ** are the original data observed twice per day and ai' are the filtered data. We denote this band-pass filter as F(I). For the wave disturbances with period from 1.5 days to 2.0 days, we use the following band-pass filter We denote this band-pass filter as F(II). Response curves of the band-pass filter are shown in Fig. 2. Fig. 1. Upper air observation stations over the China continent and the Eastern China Sea in February 1968. The enclosed areas are used in computation of vertical velocity and heat source. 3. Wave disturbance with 4-to 5-day period 3.1. Mean flow and wave disturbances Fig. 3 shows the latitude-height cross section of the mean zonal winds at the longitude about 117 *E. Weak easterly currents exist near the surface and the maximum westerlies appear at

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 13 Fig. 3. Latitude-height cross sections of the mean zonal winds along the longitude about 117*E. 300mb at latitudes about 28*N. The position of the maximum westerlies moves a little northward in the central region of the Eastern China Sea (not shown). Fig. 4 illustrates the longitude-time cross section of wind at 850mb along the latitudes between 24*N and 26*N from 12 to 28 February. It can be noted that the fluctuations of the north-sout h wind direction propagate eastward from the Chin continent to the Eastern China Sea region wit. 4- to 5-day period. 3.2. Power spectra of various parameters In order to examine the fluctuations of the disturbances more clearly, we obtain the power spectra of u (eastward wind), * (northward wind ), T (temperature), R (mixing ratio of the water vapor) and * (geopotential height). Since the data in the continent have many missing data we compensate the missing data by linear interpolation. We smooth the data to remove Fig. 4. Longitude-time cross sections of the wind between 24*N and 26*N at 850 mb from 12 to 28 February 1968. Area of southerly winds are shaded. the fluctuations with period shorter than one day. We take 24 (12 days) as a maximum lag number. Fig. 5 shows the vertical distributions of the powerspectra of * and T at three stations which lie along latitudes about 24*N. Large spectral densities exist in both the spectra of * and T in the period range longer than 10 days. However, since the total period of the time series data is 29 days, the resolution of the spectral density is not fine enough in the period range longer than 10 days. We shall thus pay attention to the disturbances with period shorter than 10 days.

14 Journal of the Meteorological Society of Japan Vol. 51, No. 1 Fig. 5 (a) Fig. 5 (b) Fig. 5 (c) Fig. 5. Vertical distributions of the power spectra of north-south wind and temperature at three stations. As for the v-spectra, a weak peak exists at 4- to 5-day period in the lower troposphere below 500mb at stations in the continent and this peak becomes remarkable at Ishigakijima. At Ishigakijima the spectral density is large also in the upper troposphere centered about 300mb. As for the T-spectra, there are no spectral peaks in the period range shorter than 10 days at Nanning but a weak peak appears at 4- to 5-day period at 850mb at Shantou and the spectral density in this period range becomes large at levels centered about 800mb at Ishigakijima. The spectral peaks appearing in the power spectra of and T may be due to the disturbances as showin in Fig. 4. The power spectra of other parameters have no pronounced peaks in the 4- to 5-day period Fig. 6. Vertical distributions of the power spectra of east-west, geopotential height and mixing ratio at Ishigakijima. range at stations in the continent (not shown) but have spectral peaks over the sea. Fig. 6 illustrates the power spectra of u, * and R at Ishigakijima. As for the u-spectra, large power densities exist in the middle troposphere centered about 400mb in the 4- to 5-day period range. Both in the *-spectra and the R-spectra, there are spectral peaks in this period range in the lower layers. Fig. 7 shows the horizontal distribution of the amplitude of the *-component integrated from 4.0 day to 4.8 day periods at 850mb. Spectral peaks in this period range exist at stations to the south of 32*N but do not exist at higher latitudes. The amplitude over the sea is larger than that over the continent. In the continent, the amplitude decreases with increase in latitude.

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 15 Fig. 7. Horizontal distributions of the amplitude of the v-component due to the disturbances with periods from 4 days to 4.8 days at 850mb. Double circles denote that there exist spectral peaks in the v- component in these period range. Fig. 8. Relation between the phase difference (**) of the v-component and the longitudinal difference of the stations (**) for 4.0-6.0 day period at levels from 1000 to 700mb. Large circles denote mb results whose coherence are larger than 95% probability level and smal small circles denote results whose coherence are ranging between 95 %-80 % probability level. 3.3. Longitudinal scale From the phase difference of the v-component between various stations, we can obtain the phase speed and the longitudinal scale of the disturbances. Fig. 8 shows the relation between phase difference (**) and longitudinal separation of stations (**) at levels from 1,000mb to 700mb. Fig. 9. Schematic structure of the disturbasces with the 4.0-6.0 day period. * denote trough and ridge axes estimated from the v-component, * warm and cold regions and * maximum moisture MA h. A M P We use 6 stations which lie between 22*N and 27*N and take 15 pairs of stations. Results lie on the linear line and show that the disturbances propagate eastward. The wavelength estimated from the inclination of the linear line is about 4,500km. 3.4. Vertical structure In this sub-section we shall examine vertical structure of the disturbance from the cross spectrum between various parameters. We calculate the vertical phase difference of the v-component on the basis of 850mb and the phase difference between various parameters (v, T and R) on the basis of the v-component at each station. Fig. 9 shows the phase relation between v, T and R averaged at stations in the Eastern China Sea. In the lower troposphere below 500mb, the tilt of the trough axis is westward with height and warm and moist air lies to the east of the trough. These structures are almost similar to those of the baroclinic wave. In the upper troposphere, the trough axis tilts eastsward with height and temperature distribution is out of phase with that in the lower troposphere. However, it is not clear how the upper-level and low-level disturbances are vertically coupled. At the China continent, the amplitude of the disturbance is small and the coherences between various parameters are also small. It can be noted, however, that the vertical structure of the disturbance is almost equal to that obtained in

16 Journal of the Meteorological Society of Japan Vol. 51, No. 1 d*/ dt from the following equation Fig. 10. Power spectra of * at 925mb and * at 700mb in three blokes. the Eastern China Sea and the vertical tilt of the trough axis is smaller than that in the sea (not shown). Averaged structure containing vertical velocity and heat source in each block as shown in Fig. 1 will be obtained by the use of the band-pass filter in the next subsection. 3.5. Computation of vertical velocity and the structure of the disturbance The vertical velocity and the heat source were computed by means of the so-called "direct method". The procedure of calculation is similar to that used in Nitta (1972). The vertical p- velocity w-dp/*p is computed kinematically by the use of continuity relation which is integrated over the three blocks enclosed by thick dashed lines in Fig. 1. We assume that w vanishes at 1,000mb. Since there are many missing data in the upper troposphere above 500mb, we obtain * at levels below 500mb. In addition to the calculation of the vertical velocity. we obfain the relative vorticity * in each block. In order to estimate the heat source, we calculate the individual change of potential temperature where cp is the specific heat of air under constant pressure, * = cp/cv, cv is the specific heat of air under constant volume and over-bars denote areal mean under considerations. Fig. 10. illustarates the power spectra of the relative vorticity * at 925mb and * at 700mb for three blocks. The power density of of block I is very small and has no spectral peaks. n the contrary, there exists large power density O of *in the period range about 4 days in blocks II and III and the power density of block III is larger than that of block II. There are no spectral peaks in the period range about 4 days in w-spectra for blocks I and II but the peaks exist for block III. Recently Ninomiya (1972) has computed the vertical velocity in the area which closely agrees with block III in this study and obtained spectral peaks of the vertical velocity in the 4-day period range. The result of the w-spectra of block III is nearly equal to that obtained by Ninomiya (1972). The results about the *-spectra and the *-spectra in three blocks indicate that the disturbance with 4-day period has small amplitude in the continent and its amplitude becomes large with propagating eastward. Next, we obtain the composite structure of the disturbance in three blocks from the filtered data of v, T, * and Q1. In order to pick up the disturbances with 4- to 5-day period, we apply the band-pass filter F (I) described in section 2 to time series data. Since these disturbances do not appear clearly during the period from 1 to 10 February and the data during 2 days are cut at the end of the analyzed period by the filtered method, we pick up four disturbances existing during the period from 11 to 27 February. We use the point where the wind direction at 850mb changes from the northward direction to the southwarcd direction as the central figure and compose the filtered data of the four disturbances. Fig. 11 shows the composite structure of the disturbance in three blocks. The amplitudes of v, T, * and Q1 become large as the wave

February 1973 Tsuyoshi Nitta, Masatoshi Nanbn and Masanori Yashizaki 17 Fig. 11(a) propagates eastward. The trough axes tilt westward with altitude and the tilt becomes large as the wave moves eastward. Warm air lies to the east of the trough and its maximum amplitude exists near 850mb. The upward motion also lies to the east of the trough. In block I, the diabatic heating is small but in block II, the large heating exists to the east of the trough. In block III: there exists the large heating to the east of the trough at the upper levels above 850mb and also to the west of the trough at the lower levels. The large heating at the lower layer may be due to the large supply of the sensible heat from the sea surface because the temperature is coldest to the west of the trough. The composite Fig. 11(b) structures of v and T are equal to those obtained from the spectral method. 3.6. Development of disturbances As mentioned in the previous sub-section, the averaged amplitude of the four disturbances increases when they propagate eastward especially when they enter the sea regions. All disturbances, however, do not develop. Fig. 12 shows the amplitude of v at 850mb, T at 850mb and o at 700mb of each disturbance in three blocks. The amplitude is estimated from the time-filtered data. Wave-(2) develops suddenly and wave-(3) amplifies normally. Wave-(1) has small amplitude and develops slowly. On the contrary, swave-(4)

18 Journal of the Meteorological Society of Japan Vol. 51, No. 1 Fig. 11(c) Fig. 11(d) Fig. 11. Composite structure of the disturbances with period about 4 days picked up by the band-pass filter. (a) north-south wind (b) temperature (c) vertical p-velocity (d) diabatic heating rate. amplifies while it is propagating from block I along about 28*N from 12 GMT 12 to 12 GMT to block II, but damps from block II to block 14 February. Wind, T and R are obtained by III. Then we shall examing case studies of two averaging the values at a nothern station and a disturbances, i.e., the one is the case of the southern station. T is the deviation from the wave-(2) as the developing case and the other is averaged time-mean value of eight stations. * the case of the wave-(4) as the damping case. and Q1 are computed in the triangular regions Also careful study will be made about the by dividing each block into two parts and are mechanism of the development according to the averaged over two neighboring regions because difference of the two cases. the computed values over the triangular regions Fig. 13 illustrated longitude-height cross are changeable. A strong trough at 500mb sections of wind, T,R, * and Ql for the wave-(2) appeats over the Himalayas on 11 February (not

February 1973 Tsuyashi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 19 Fig. 12. The amplitude of v at 850mb, T at 850mb and * at 700mb of the four disturbances in three blocks. Fig. 13. Longitude-height cross sections of the wind, temperature, mixing ratio, vertical velocity and diabatic heating from 12 GMT 12 to 12 GMT 14 February. shown) and propagates eastward. On 12 GMT 12 February, the trough at 500mb exists at about 110*E and the weak upward motion exists to the east of the trough. Relatively large mixing ratio lies over the Eastern China Sea region but the temperature deviation is small. On 12 GMT 13, the trough line extends down to the surface, southerly winds flow in the lower troposphere and the warm air appears to the east of the trough. The amplitude of upward motion does not become so large but the region of upward motion moves eastward and enters the sea area. On 12 GMT 14, disturbance develops extensively. Upward motion and the

20 Journal of the Meteorological Society of Japan Vol. 51, No. Fig. 14. The same as Fig. 14 except from 00 GMT 22 to 00 GMT 23 February. Fig. 15. Isolines of the height at 500mb during the time when the lower tropospheric disturbances enter the Eastern China Sea in the four cases. (a) Wave-(1) (b) Wave-(2) (c) Wave-(3) (d) Wave-(4).

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 21 diabatic heating become larger. This disturbance develops greatly during the time when the upward motion enters the sea region. Probably the diabatic heating over the sea play an important role in the development of this disturbance. Fig. 14 shows longitude-height cross sections from 00 GMT 22 to 00 GMT 23 for the wave- (4). On 00 GMT 22, a weak disturbance exists at 850mb at about 118 *E and the upward motion occurs to the east of the disturbance. Both the trough at 500mb and the warm core to the east, of the disturbance do not exist in this case. At the next time step, the disturbance propagates eastward and both the upward motion and the diabatic heating intensify. But the southerly wind is weak and the warm core to the east of the disturbance does not exist. On 00 GMT 23, the disturbance of the wind disappears and the upward motion and the diabatic heating decay. The difference of the wave-(4) from the wave- (2) is that both the trough at 500mb and the warm core to the east of the disturbance do not appear and the southerly wind in the lower troposphere is very weak. Results of ther developing cases (figures are not shown) show the existence of the strong southerly wind and the warm core to the east of the disturbances during the time when they enter the sea region. The relation between the development of the disturbance and the trough in the upper troposphere has been discussed by many authors. Fig. 15 illustrates isolines of the height at 500mb when the Eastern China Sea. As for the developing cases (wave-(2) and wave-(3)), the disturbances seem to couple with upper troughs. In the case of wave-(2) as described before, the strong trough appears in the southern part of the Himalayas and propagates eastward, In the case of wave-(3), the trough of the northern region elongates to the southern region and the disturbance appears below the upper trough and develops. In the cases of wave-(4) which is the decaying case, the upper trough does not exist above the lower disturbance. When the lower disturbance enters the Eastern China Sea, the northern trough at 500mb lies to the east of the lower disturbance (see Fig. 15 (d)) and then the disturbance decays. The strong southerly wind and the warm core in the lower troposphere may be produced by the upper trough. Therefore the existence of the upper trough during the Fig. 16. Time-height cross sections of the winds at Minamidaitojima from 3 to 8 February 1968' time when the disturbance enters the sea is important for the development of the wave as well as the diabatic heating. Detailed analyses of the diabatic heating over the sea and the coupling between the lower disturbances and the upper troughs must be studied in future. 4. Wave disturbance with 1.5- to 2-day period 4.1. Time section and power spectra of the wind The 4- to 5-day period disturbances appear during the period starting 10 February as described in the previous section. On the contrary, the short period disturbances appear during the period from 1 to 10 February. Fig. 16 shows the time-height section of the wind at Minamidaitojima from 03 GMT 3 to 21 GMT 8. It can be noted that the short period disturbances exist in the lower troposphere below 600mb. The period is about 2 days. In order to detect the disturbance more clearly, we take power spectra of the wind. Fig. 17 denotes the power spectra of the v-component at Ishigakijima (N 24*, E 124*), Kadena (N 26*, E128*), Naze (N28*, E130*) and Minamidaitojima (N26*, E121 *). The wind data at Ishigakijima, Naze and Minamidaitojima are observed four times a day and the data at Kadena are available twice a day. We take 6 days as the maximum lag period. We can notice that in addition to the large power density of the 4-day period, rather large power density exists in the period range from 1.5 days to 2 days in the lower troposphere centered about 900mb at the stations except Naze. The spectral density due to the short period disturbances is about a half in comparison with that due to the 4- to 5-day period disturbances. Indistinct spectral peaks at Naze may be due to the northern location of

22 Journal of the Meteorological Society of Japan Vol. 51. No. I Fig. 17. Vertical distributions of the v-spectra at four stations. the station with respect to the center of the disturbance. In the power spectra of the other parameters, there are no distinct spectral peaks in the period range shorter than 4 days. 4.2. Computation of the vertical velocity In order to obtain the vertical motion of disturbances with the short wavelength, we use the triangle block as shown in Fig. 1 denoted by block IV. The length of oneside is about 300km. Since the data of Kadena is obtained twice a day, computation is also made twice a day. The wind data are available from the surface up to 200mb and the vertical velocity is obtained by the same manner as in the previous section. However, since the error of the wind

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 23 Fig. 18. Vertical distributions of the *-spectra in block IV. Fig. 20. Vertical phase difference of the v-component measured from 850 mb at four stations. Fig. 19. Relation between the phase difference data accumulates, it is needed to modify the computed vertical velocity to insure that * in the upper troposphere becomes small. If we assume (1,000mb) =0, * (200mb) is obtained * as follows ; Since the error of * V becomes large with increase in the altitude, we modify the divergence as where (**V)* is the modified divergence. Thus the integrated value of the modified divergence from 1,000mb to 200mb vanishes. The power spectra of * in block IV are shown in Fig. 18. In acdition to the spectral peak near the 4-day period, another spectral peak exists near the 2-day period. This shorter period spectral peak may correspond to that in the v-spectra. The power density due to the short period disturbance becomes maximum at 600mb and its value is larger than that of the 4-day period disturbance. 4.3. Structure of the disturbance We estimate the phase velocity and the longitudinal wavelength of the disturbance from the longitudinal phase difference of the v-component. Fig. 19 shows the phase difference of the v- component between two stations at levels from the surface to 800mb. We dealt with the data at four stations. The phase difference estimated from the data observed twice per day (triangles in Fig. 19) is somewhat smaller than that from

24 Journal of the Meteorological Society of Japan Vol. 51, No. 1 Fig. 21(a) Fig. 21(c) Fig. 21. Composite structure of the disturbances with `period from 1.5 days to 2.0 days picked up by the band-pass filter. (a) v- component -velocity (b) temperature (c) vertical p the data observed four times per day (circles in Fig. 19). From the result of (**-**) diagram, the disturbance propagates eastward and its wavelength is about 2.000km. The vertical phase difference of the v-component on the basis of 850mb is illustrated in Fig. 20. It is noted that the disturbance is recognized mostly in the lower troposphere below 600mb and the wave axis tilts slightly westward with increase in altitude. Next let us obtain the composite structure of the disturbance by applying the band-pass filter F(II) described in section 2 to the time series data averaged in block IV. Since these distur- Fig. 21(b) bances appear distinctly during the period from 3 to 8 February, we pick up three disturbances which appear during that period and construct the vertical structure of the disturbance. Fig. 21. (a) shows the vertical distribution of the v-component. The maximum amplitude takes place at levels from 900mb to 850mb with the value of about 4 m/s. The trough lines tilt slightly westward with increase in altitude. This is quite coincidental with that obtained from the power spectra. Fig. 21 (b) illustrates the vertical distribution of the temperature. There exists warm air at the trough and to the east of it. Maximum temperature exists at about 850mb with the value of about 0.8*. The vertical distribution of the upward motion is shown in Fig. 21(c). Upward motion exists to the east of the trough. In spite of the shortness of wavelength, these structures of the disturbances are similar to those of the baroclinic waves. 4.4. Synoptic pattern The synoptic patterns at the surface and 850mb on 12 GMT 6 February are shown in Fig. 22. We can notice that two waves exist at 127*E and 144*E at both levels and warm air flows into the trough region. The wavelength of these disturbances is about 2,000km. The disturbances do not appear at the upper levels above 700mb, although patterns are not shown here. The centers of the cyclones at 850mb lie slightly to the east of those at the surface and these vertical tilts differ from those obtained by the spectral method. Since the block IV lies to the north of the center of the disturbances,

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 25 Fig. 22. Synoptic patterns at the surface and 850mb on 12 GMT 6 February. Solid lines denote isobars at the surface and isolines of the geopotential height at 850mb and dashed lines denote isotherms. the position of the surface trough estimated from the v-component of the block IV may be rather east to that of the pressure trough due to the surface friction. Fig. 23 illustrates the longitudeheight cross section of the temperature deviations along 23*N from 120*E to 150*E. Warm air invades into the east of the trough at lower levels below 850mb and the large scale temperature contrast exists at upper levels above 700mb. This temperature contrast at upper levels relates

26 Journal of the Meteorological Society of Japan Vol. 51, No. 1 Cable 1. Mean zonal wind, temperature and Fig. 23. Longitude-height cross section of the temperature deviated from time and longitudinal mean temperature along 23*N on 12 GMT 6 February. L denote the positions on the surface cyclones. to the cyclone in the northern latitudes. The disturbances appearing in the synloptic patterns correspond to the short period disturbances obtained from the time-series data. 4.5. Mechanism of development Although the disturbance has the short wavelength, the structure is similar to that of a usual baroclinic wave. Tokioka (1970) investigated the non-geostrophic instability of * baoclinic fluid and showed that the wavelength of the baroclinic wave becomes smaller as the Richardson number of the fluid decreases. We estimate the Richardson number defined by from the monthly mean data averaged over block IV for the month of February. Table 1 shows the Richardson number at constant pressure levels. The value of Ri is smaller than 6 at the levels lower than 400mb. Therefore we can conclude that the disturbance with the short wavelength is a baroclinic wave which develops in the atmosphere with the small Richardson number. 6. Conclusions and remarks From the spectral analysis and the dynamical computation of the vertical velocity and the diabatic heating, we can obtain two types of disturbances. The one is the long wave with the period of 4*5 days and the wavelength of 4,000 5,000km and the other is the medium-scale * disturbances with the period of 1.5*2 days and the wavelength of 2.000km. The structure of the long wave indicates that the trough axes tilt westward with increase in altitude and warm and upward air lies to the east, of the trough. This disturbance is generated due to the baroclinicity in the southern jet currents over the China continent and amplifies as the wave moves over the Eastern China Sea. The diabatic heating from the surface and the coupling between the wave and the upper trough play an important role in marked developing cases. Detailed case studies about sudden development of the disturbance in the Eastern China Sea is desired in future. The medium-scale disturbances predominate in the southern part of the Eastern China Sea, especially during the period from 1 to 10 February. These disturbances also show structure of the baroclinic wave. Gambo (1976, a,b) and Tokioka (1970) have indicated theoretically that the wavelength of the baroclinic unstable wave decreases with decrease in Richardson number. The estimated value of Richardson number in this study is less than 6 in the lower troposphere. Since the vertical structure of the medium-scale disturbances in this study generally agrees with that of baroclinic waves, these medium-scale disturbances may be considered as the baroclinic waves which develop in the atmosphere where

February 1973 Tsuyoshi Nitta, Masatoshi Nanbu and Masanori Yoshizaki 27 the Richardson number is small. Orlanski (1968 found a medium-scale disturbance which develop; due to both effects of baroclinic instability ant Helmholtz instability when Ri,= 5 and Ro is abou 0.3. However, we can not compare Ri in thi; study directly with Ri of Orlanski (1968) becausf his model is a simplified layer model. Recently Tokioka (1972) showed that the other types o unstable waves have the greatest growth rate when the Richardson number is order of unit in his theoretical model which includes the effect: of the moist connective motions. More detailed analyses including energy budget will be needed to clarify whether other types of medium-scale disturbances which are different from baroclinic waves exist or not in the real atmosphere. Acknowledgments medium scale disturbances in the atmosphere (I). J. Meteor. Soc. Japan, 48, 173-184.., 1970b: The characteristic - feature of medium scale disturbances in the atmosphere (II). J. Meteor. Soc. Japan, 48, 315-330. Maruyama, T., 1968: Time sequence of power spectra of disturbances in the equatorial lower stratosphere in relation to the quasi-biennial oscillation. r J. Meteor. Soc. Japan, 46, 3x7-342. Matsumoto, S., S. Yoshizumi, and M. Takeuchi, 1970: On the structure of the "Bain Front" and the associated intermediate-scale disturbances in l the lower atmospohere. J. Meteor. Soc. Japan, 48, 479-491. Ninomiya, K., 1972: Heat and water-vapor budget over the East China Sea in winter season-with special emphasis on the relation among the supply from the sea surface, the convective transfer and the synoptic weather situation. J. Meteor. Soc. Japan, 50, 1-17. The authors wish to express their thanks to ' Nitta, Tsuyoshi, 1972: Energy budget of wave disturbances Professor K. Gambo for his suggestions ant: over the Marshall Islands during the years encouragement. The authors also thank Drs. T of 1956 and 1958. J. Meteor. Soc. Japan, 50, Matsuno, K. Ninomiya and T. Maruyama fog 71-84. their useful suggestions and comments. Thank: Orlanski, I., 1968: Instability of frontal waves. J. are extended to Miss S. Lovell for typing the Atmos. Sci., 25, 178-200. manuscript. This research was financially supported Tokioka, T., 1970: Non-geostrophic and non-hydrostatic stability of a baroclinic fluid. J. Meteor. in part by Funds for Scientific Research from Soc. Japan, 48, 503-520. the Ministry of Education. Tokioka, T., 1971: Supplement to non-geostrophic References Eliasen, E., 1960: On the initial development o f and non-hydrostatic stability of a baroclinic fluid and medium-scale disturbances on the fronts. J. Meteor. Soc. Japan, 49, 129-132. frontal waves. Publ. Danske Meteor. Inst., No. 13 Tokioda, T., 1972: A stability study of medium-scale 107 pp. disturbances with inclusion of convective effects. Gambo K., 1970 a: The characteristic feature o J. Meteor. Soc. Japan, 51,